The Ragone plots guided sizing of hybrid storage system for taming the wind power

Electrical Power and Energy Systems 65 (2015) 246–253

Contents lists available at ScienceDirect

Electrical Power and Energy Systems journal homepage: www.elsevier.com/locate/ijepes

The Ragone plots guided sizing of hybrid storage system for taming the wind power Yiming Zhang a, Xisheng Tang b,⇑, Zhiping Qi b, Zhaoping Liu a a b

Ningbo Institute of Material Technology & Engineering, Chinese Academy of Sciences, No. 1219 Zhongguan West Road, Zhenhai District, Ningbo, Zhejiang Province 315201, PR China Institute of Electrical Engineering, Chinese Academy of Sciences, No. 6 Beiertiao, Zhongguancun, Beijing 100190, PR China

a r t i c l e

i n f o

Article history: Received 15 September 2013 Received in revised form 27 June 2014 Accepted 12 October 2014

Keywords: Hybrid electrical energy storage Ragone plot Life cycle cost Wind power Sizing strategy

a b s t r a c t The fluctuant behaviors of wind power, and so as their effects on the stability of grid, span across different time scales. Hence, single type of electrical energy storage (EES) cannot level the fluctuation effectively. The high energy density sources (e.g. lead-acid batteries) and high power density sources (e.g. supercapacitors) are complementary in merits such as power density and energy density. As a result, the employment of the hybrid EES (HEES) is hoped to level wind power output more effectively. This work proposes a sizing strategy for determining the capacity allocation between the high energy density sources and high power density sources. The traits of this employed strategy are the introduction of energy-power relationships (Ragone plots) of EES as constraints and taking of the minimization of life cycle cost (LCC) of HESS as objective function; which respectively considers the power and energy storage characteristics of EES integrally, and reflects the economic need of renewable energy integration. The analytical process of the sizing strategy is described in detail. A case study for an analysis of specific wind power output, with the types of EES we adopted in project, is illustrated graphically based on the developed programming software platform. Ó 2014 Elsevier Ltd. All rights reserved.

Introduction As the increasing prosperity and economic growth, the capability to access to clean and reliable energy has been treated as a cornerstone of the world [1]. This requirement is more demanding especially in developing nations, such as China, which has sawn an extended period of double-digit annual increases in economic growth and energy consumption. With the wind power estimated of 3000 GW or so, available for covering nearly all of the electricity demand [2], China is increasingly swinging around and embracing this kind of clean energy [3] However, the common challenges

Abbreviations: EES, electrical energy storage; HESS, hybrid electrical energy storage; LCC, life cycle cost; ELoadmax, the maximum value of the energy that the electrical energy storage needs to undertake; Eq, the energy density (with unit of J kg1); mb, amount of the lead-acid batteries that need to be employed; msc, amount of the supercapacitors that need to be employed; PoB, unit price of leadacid batteries; PoSC, unit price of the supercapacitors; PLevel(t), power needs to be taken by hybrid electrical energy storage; Pb(t), power demands of lead-acid batteries; Psc(t), power demands of supercapacitors; PLoadmax, the maximum value of the power that the electrical eneroogy storage needs to undertake; Pq, the power density (with unit of W kg1); s, the filtering time constant. ⇑ Corresponding author. Tel.: +86 (0)10 82547108. E-mail address: [email protected] (X. Tang). http://dx.doi.org/10.1016/j.ijepes.2014.10.006 0142-0615/Ó 2014 Elsevier Ltd. All rights reserved.

facing the employment of wind power are cost and the ability of integrating this kind of renewable source into the grid. The situation of far from existing high capacity transmission lines and their difficulties in matching the electricity production with demand, as well as the variable and unpredictable characteristics of wind power, the before mentioned high wind potential cannot be utilized completely for electricity production. Currently, there are several options existed that can be implemented in order to obtain a larger wind energy contribution, including: (1) Grid reinforcement (i.e. transmission upgrades); (2) Fossil fuel utilities dispatch; and (3) Employment of energy buffer (i.e. energy storage). Grid reinforcement expands the capacity of the grid through increasing the cross section of the cables, which is usually done from erecting a new line parallel to the existing line for some part of the distance. However, this approach can be very costly and sometimes impossible due to planning restrictions [4]. Fossil fuel utilities dispatch is the measure that uses slow responding base and intermediate load generators with fast responding peak load generators to capture the chaotic behavior of wind sources [5]. However, there are increased costs associated with the additional needs for providing short-term ramping (for fulfilling frequency regulation) and hourly ramping (for satisfying load following requirements) [6,7], as well as the less efficiently operate

Y. Zhang et al. / Electrical Power and Energy Systems 65 (2015) 246–253

results from having a suboptimal mix of units online because of errors in the wind forecast [8,9]. Electrical energy storage (EES) is another choice that has been proposed to enable the greater use of wind power through leveling its fluctuations; and is characterized in terms of its fast response over diverse timescales and control flexibility [8,10,11]. Compared with the grid reinforcement and fossil fuel utilities dispatch, EES employment solutions are likely to be more economical and technical feasible [12–14]. Currently, there are numerous literatures reviewing and discussing the employment of suitable storage technologies for wind power integration [5,15–19]. Owning to the fluctuation characteristics of wind speed and direction, it would result in the random combinations of power fluctuations with time scales ranging from cycles of several tens of seconds, cycles of several minutes, to cycles of several tens of minutes and cycles of hours [20]. Meanwhile, each type of EES has its own traits (i.e. response characteristics, storage capacity, cycle life, capital cost) and so the preferred niche applications. As a result, employing single type of EES for steering wind power cannot technologically or economically satisfies the power system requirements. The high energy density sources (e.g. lead-acid battery, lithium-ion battery) and high power density sources (e.g. supercapacitor, flywheel) are highly complementary in power density, energy density and life cycle. The employment of the hybrid of these two types of EES is hoped to have capability for leveling wind power output effectively and economically. Until now, there have been plenty of works being tried to employ hybrid (or composite) energy storage, accompanying with relevant devices and suitable charging/discharging strategies, to mitigate the intermittent characteristic of wind power and the like renewable [21–35]. Here, a strategic question arisen is how large the hybrid EES (HEES) capacity should be assigned; as well as how to distribute the total capacity, that the HEES needs to undertake, between the hybrid sources in order to fulfill the in-grid requirement economically. Currently, Li et al. [36] used one type of hybrid energy storage system, composing of supercapacitor and lithium-ion battery, to enhance wind power predictability; and so the grid stability. Within their work, an artificial neural network (ANN) based control algorithm was proposed to control the input/output of the storage system, in order to ensure the rated storage system capacity could satisfy the wind farm’s running life. Chen et al. [37] employed one type of HEES including batteries and supercapacitors to smooth the fluctuation of wind power. They firstly based on the predictive output power of wind farm, and employ particle swarm optimization (PSO) algorithm to derive the desired power schedule of wind farm. From that the desired power compensation of the HEES can be derived. The PSO algorithm was implemented again to calculate the power contribution from the battery, in which reducing the state transitions of the battery is considered in order to extend the lifetime of the battery. The rest part of power compensation of HEES, contributed by supercapacitor, can be derived. However, both of these studies stood at the point that the employment of supercapacitors is aimed to prevent the batteries from frequent charge/discharge state transitions; and so the capacity of employed batteries can be decreased, or the lifetime can be extended. It is worth noting that the power/energy mutual constraints of the supercapacitors, as well as of the batteries, are not taken into account. Furthermore, it is well known that the price of supercapacitors are always much higher than that of the batteries; as a result, employing supercapacitors to ‘protect’ the batteries not necessarily reducing the total cost of the HEES. In this study, we propose a strategy that applying Ragone plots to allocate the capacity within HEES. Ragone plot is the curve that displays the energy available to load as a function of the power, which differentiate energy storage devices by means of the available energy and power [38]. As mentioned by Christen and Ohler [39], this kind of

247

method has a two-fold advantage for EES optimization including rigorously defined for any kind of EES [40] and readily display the two parameters with cost impact. System formulation Stems from the stochastic nature of the wind speed, the total captured wind power is highly time-varying. Fig. 1 shows the one-day wind power output profile of a medium size wind farm (11 MW capacities) in China, whose fluctuation rate reaches 1.9 MW min1; and would severely affect the stability of the power system. Due to that reason, system operators sometimes set instructions of power variation to wind farms, and demand them to follow the defined generator profiles (i.e. desired power schedule). A schematic diagram of the studied system is shown in Fig. 2, which depicts the employment of HEES for mitigating wind power fluctuations and variability. The batteries (lead-acid) and supercapacitors are together connected to the grid at the common coupling point, which are charged and discharged through bidirectional DC/AC power converters to level the wind power for satisfying the specified generator profiles. Here, we will use the historical data to estimate the required capacity assignment; and the following section will present the detailed sizing strategy that would hopefully enable the wind farm outputs to meet the power system requirements, with minimum investment. HEES sizing strategy The optimized sizing of HEES, in this work, treats the capacity allocation of lead-acid batteries and supercapacitors, with the fulfillment of enabling the wind farm power output to follow the defined generator profiles. As said by Irving and Song [41], one optimization problem is a kind of mathematical model where to minimize numerical values that represent something undesirable, or maximize something which is desirable, under certain constraints. In this study, due to ‘‘. . .this often-characterized ‘need’ for energy storage to enable renewable integration is actually an economic question’’, as said by Denholm et al. [8], we take the minimization of life cycle cost (LCC) of the hybrid EES as the objective for this optimization process, while considering the leveling requirement and technical characteristics of both types of EES (viz. leadacid batteries and supercapacitors) as constraints. The following sections of this part entail the objective function and constraints, as well as the optimization strategy.

Fig. 1. Daily wind power output profile of one wind farm in China.

248

Y. Zhang et al. / Electrical Power and Energy Systems 65 (2015) 246–253

In conclusion, the objective function can be expressed as: PWF

mb  PoB þ msc  PoSC

Wind Farm (WF)

ð2Þ

The Constraints PTotal

Batteries

Grid

The constraints include both the grid requirement, i.e. leveling requirement, and also the technical constraints of EES. The detailed analysis is as follow.

DC/AC

PHEES

Leveling requirement The HEES consisting of lead-acid batteries and supercapacitors is required to mitigate the wind power fluctuations for following the defined power generation profile; and so the first constraint is represented as:

Supercapacitors

DC/AC

Hybrid EES (HEES)

Pb ðtÞ þ Psc ðtÞ ¼ PLev el ðtÞ

Fig. 2. Integration of hybrid electrical energy storage into the wind power generators.

Eρ (Jkg-1)

where Pb(t) and Psc(t) are the power demands of lead-acid batteries and supercapacitors at time t; while PLevel(t) is the power that the HEES needs to undertake. Technical constraints of EES As mentioned in Section ‘Introduction’, the Ragone plot depicts the mutual restriction between the rate capability (power) and storage capacity (energy) of EES. As a result, we specify the constraints of employed two types of EES (i.e. lead-acid batteries and supercapacitors) by employing their mathematical expressions of Ragone plots, which integrally consider both of the power and energy storage characteristics of EES. These constraints can be expressed as:

ELoadmax= TLmin×PLoadmax Pρ , Eρ

Eρ= Rag(Pρ)

Pρ (Wkg-1) O Fig. 3. The determination of power density (Pq) and energy density (Eq) for electrical energy storage for the employment using Ragone plot.

The objective function – life cycle cost (LCC) The life cycle cost (LCC) includes the costs that over the entire life cycle of a product or process, which always encompass the costs such as initial investment, capital, operating and maintenance [42]. In this case, the LCC is defined as embracing EES Purchasing Expense and Other Expense (i.e., LCC = Purchasing Expense + Other Expense), while the Other Expense part contains ‘‘control and accessory cost’’ (Cca), ‘‘auxiliary and adjusting cost’’ (Caa), ‘‘recycle and environment spending’’ (Cre) and ‘‘operation and maintenance cost’’ (Com). Here, we assume above mentioned four elements within the Other Expense part can be estimated and given directly by experienced constructors; and the Purchasing Expense is related to the amount of EES employed (Eq. (1)), which need to be calculated and optimized toward the named objective under the given constraints. It is noteworthy that in reality the Other Expense part may also be some functions of the EES amount, and need to be evaluated. In this work, we are mainly focus on the detailed calculation of Purchasing Expense.

Purchasing Expense ¼ mb  PoB þ msc  PoSC

ð3Þ

ð1Þ

where mb and msc are the amount of the lead-acid batteries and supercapacitors that need to be employed in the hybrid EES, respectively; and PoB and PoSC are the unit price of the lead-acid batteries and supercapacitors, which can be given directly.

Eq-battery ¼ RagðPq-battery Þ

ð4Þ

Eq-supercapacitor ¼ RagðPq-supercapacitor Þ

ð5Þ

The above two equations express the energy densities as functions of power densities. The Eq-battery and Pq-battery are the energy density and power density for the lead-acid battery respectively; while Eq-supercapacitor and Pq-supercapacitor are the energy density and power density for the supercapacitor respectively. The detailed mathematical expression of the Ragone plots for the lead-acid batteries and supercapacitors that we adopted in the Case Study will be shown in Section ‘Case studies’. Summarize the above formulae, the optimization problem in this work can be formulated as:

Min:mb  PoB þ msc  PoSC Subject to : mb  P q-battery þ msc  Pq-supercapacity P PLev el ðtÞ Z T PLev el ðtÞ mb  Eq-battery þ msc  Eq-supercapacity P 0

ð6Þ

where T is the EES running time Eq-battery ¼ RagðPq-battery Þ Eq-supercapacity ¼ RagðP q-supercapacity Þ Optimized sizing strategy In this section, the process of analyzing and optimizing the capacity allocation between lead-acid batteries and supercapacitors, by using Ragone plot, is described. Considering the running characteristics of lead-acid batteries and supercapacitors, it is thought that lead-acid batteries are used to bear the low-frequency fluctuations, while supercapacitors are employed to support the high-frequency fluctuations. In this work, we adopt similar strategy that employed in [27,43] to differentiate the fluctuations with low- and high- frequencies; that is, introduce a filtering time constant to PLevel(t) data and make the fluctuations

Y. Zhang et al. / Electrical Power and Energy Systems 65 (2015) 246–253

Wind Power Historical Data

249

Defined generator Profile

Minus The Power that the HEES need to Undertake, PLevel(t) Filtering Time Constant, τ

First-order Low Pass Filter

Power Demand of Lead-acid Batteries, Pb(t)

Power Demand of Supercapacitors, Psc(t)

Maximum Value of the Power and Energy from Pb(t), PLoadmax and ELoadmax

Maximum Value of the Power and Energy from PSC(t), PLoadmax and ELoadmax

Ragone Plot of Lead-acid Batteries

Ragone Plot of Supercapacitors E

E

Draw a Line: y = P Lo a d m a x × x; Lo a d m a x Intersect the Ragone Plot for Obtaining Position of (Pρ , Eρ)

Draw a Line: y = P Lo a d m a x ×x; Lo a d m a x Intersect the Ragone Plot for Obtaining Position of (Pρ , Eρ)

Calculate the Amount of EES: mb = PLoadmax/Pρ -battery or ELoadmax/Eρ -battery

Calculate the Amount of EES: msc = PLoadmax /Pρ -supercapacitor or ELoadmax /Eρ -supercapacitor

Unit Price of the Lead-acid Batteries, PoB

Multiply

Unit Price of the Supercapacitors, PoSC

Multiply

Plus Purchasing Expense

Other Expense Plus Life cycle cost (LCC)

Fig. 4. Flow chart of the hybrid electrical energy storage capacity allocation.

differentiation with a discrete low pass filter. In this analysis, it is assumed that the lead-acid batteries and supercapacitors have no efficiency losses and they give immediate responses to the filtering suggestions. The first-order passive low pass filter can be described mathematically as [44,45]:

sY 0 þ Y ¼ X

ð7Þ

where s is the filtering time constant corresponding to the power demand of supercapacitors PSC(t); Y is the filter outputs corresponding to the power demand of lead-acid batteries, Pb(t); Y0 is the derivative of Y and X is the filter inputs corresponding to the power that the HEES need to undertake, PLevel(t). By choosing different values of s, the influence of the fluctuations differentiation on the LCC can be studied. The full steps of are shown below.  Firstly, the power that the HEES need to undertake PLevel(t) can be derived from the wind power historical data and the defined generator profile. Then the introduction of filtering time constant s will decompose the PLevel(t) into Pb(t) and Psc(t);

different choice of s will give various types of HEES sizing, which will further affect the LCC values of the HEES.  Secondly, for a given load profile (e.g. Pb(t) and Psc(t) for leadacid batteries and supercapacitors, respectively), the sizing strategy of the EES is analyzed below: it is evident that the size (including the power capability and storage capacity) of one EES is determined by satisfying the maximum value of the power and energy (PLoadmax and ELoadmax) that the EES needs to undertake. Considering the fact that the Ragone plots show the behaviors of EES that when charged/discharged with a higher power, the lower energy it can absorb/release; and so it can concluded that if the size of EES is determined for satisfying the PLoadmax and ELoadmax simultaneously, then the EES with this size can afford any instant of a load with maximum values of PLoadmax and ELoadmax.  From the above analysis, the constraints from Ragone plot are introduced in. As mentioned in Section ‘Technical constraints of EES’, the Ragone plot of one kind of EES can be represented as Eq = Rag(Pq). In order to satisfy the load requirement, as discussed above, the EES have to fulfill the capability of absorbing/

250

Y. Zhang et al. / Electrical Power and Energy Systems 65 (2015) 246–253

releasing energy with power rate of PLoadmax while having the energy amount equal to or greater than ELoadmax. This can be expressed as:

 EES ¼ mES  Eq ¼ mES  Rag

PLoadmax mES

 P ELoadmax

ð8Þ

Where EES represents the energy absorb/release of mES kg of EES with the charging/discharging power of PLoadmax. As the result, the minimum value of mES that also satisfies the technical constraints of EES can be acquired under the condition of mES  Rag (PLoadmax/mES) = ELoadmax. And so, transform the Eq. (3) could have the below form:

Rag



P Loadmax mES

P Loadmax mES

 ¼

ELoadmax PLoadmax

ð9Þ

The Eq. (4) reveals the way to obtain the (Pq, Eq) point of EES that can satisfy the PLoadmax and ELoadmax at meantime. The process of the (Pq, Eq) determination can be stated as follow (depicted in Fig. 3): drawing a line passing the original point with slope of TLim = ELoadmax/PLoadmax (where TLim can be denoted as the minimum lasting time that the EES need to bear) and intersect the Ragone plot of the EES; then the intersection is the point (Pq, Eq) that want to acquire. From that, the amount of EES that needs to be employed can be calculated by dividing the ELoadmax by Eq or dividing the PLoadmax by Pq.  From the above mentioned process, the value of the LCC can be calculated: i) by trying different values of s, the power that the HEES need to undertake PLevel(t) will be decomposed into various types of Pb(t) and Psc(t) combination; ii) For each combination of Pb(t) and Psc(t), the amount of lead-acid batteries mB and supercapacitors mSC can be calculated, respectively; iii) then from Eq. (1), the value of Purchasing Expense, and further the LCC, can be got for each Pb(t) and Psc(t) combination. The above stated process can be expressed with a flow chart, which is shown in Fig. 4.  Within the various values of LCC, the minimum value is chosen, which return back to derive the values of mB and mSC that would give minimum LCC values. Case studies These case studies attempt to determine the optimal capacity allocation between lead-acid batteries and supercapacitors, which would level the wind power fluctuation (the wind power data Pw(t) is shown in Fig. 1) both technically and economically. The three case studies show the different sizing results for choosing different defined generator profiles. In all of these cases, the defined generator profile is closer to the wind output power derived from historical data, which is aimed to reduce the capacity demand of HEES. The Ragone plots representing the lead-acid battery and supercapacitor, in this work, are derived from fitting the experimental power and energy data, which are Eq-battery = 0.0015  Pq-battery3 + 0.093  Pq-battery2 – 2.3112  Pq-battery + 33.876 (0 < Pq-battery < 36.8) and Eq-supercapacitor = 0.0062  Pq-supercapacitor + 5.8 (0 < Pq-supercapacitor < 935.48). Case 1 In this case, the defined generator profile Pd(t) is shown in Fig. 5, whose requirement of the wind power fluctuation leveling is loose. The steps of determining the capacity allocation between the lead-acid batteries and supercapacitors are shown below (following the flow chart shown in Fig. 4):

Fig. 5. Defined generator profile that requires low level of smoothing.

Step 1: Derive the power PLevel(t) that the HEES need to undertake for satisfying the defined generator profile

P Lev el ðtÞ ¼ Pw ðtÞ  Pd ðtÞ

ð10Þ

Step 2: Choose an arbitrary filter constant, say for example 2, in order to differentiate the power demand between lead-acid batteries and supercapacitors. The power undertaken by lead-acid batteries Pb(t) can be got from the MATLAB command filter (ones(1,2)/2,1, PLevel(t)), and the power undertaken by supercapacitors Psc(t) is PLevel(t)  Pb(t). Step 3: Derive the maximum value of the power and energy from Pb(t) and Psc(t) respectively. In this case, the maximum value of power for Pb(t) and Psc(t) are 4.1 MW and 1.9 MW. The maximum value of energy from Pb(t) and Psc(t) can be determined from integrating the Pb(t) and Psc(t), and then choose the maximum values. In this case, they are 1.3 MW h and 0.023 MW h for Pb(t) and Psc(t) respectively. Step 4: Draw a line y = (1.3/4.1)  x, and intersect Ragone plot of lead-acid battery, which is y = 0.0015  x3 + 0.093  x2– 2.3112  x + 33.876; and obtain position of (28, 9.0), which are the values of power density and energy density for the leadacid battery in this differentiation (choose filter constant as 2). Step 5: Similarly, draw a line y = (0.023/1.9)  x, and intersect Ragone plot of lead-acid battery, which is y = 0.0062  x + 5.8; and obtain position of (317, 3.84), which are the values of power density and energy density for the supercapacitor in this differentiation (choose filter constant as 2). Step 6: Calculate the amount of lead-acid batteries and supercapacitors need to allocate from equations: mb = PLoadmax/Pq-battery (or mb = ELoadmax/Eq-battery) and msc = PLoadmax/Pq-supercapacitor (or msc = ELoadmax/Eq-supercapacitor). The mb and msc are 146429 kg and 5993.69 kg, respectively. Step 7: Calculate the Purchase Expense. The unit price of the lead-acid batteries (PoB) and supercapacitors (PoSC) are introduced in, which are ¥35/kg for lead-acid batteries and ¥500/ kg for supercapacitors. The Purchase Expense is calculated as ¥ 8,121,860. Step 8: Employ trial-and-error method to choose another different filter constant; then, repeat Step 2 to Step7 to calculate the Purchase Expense for different choice of filter constants. Step 9: Finally, from comparing the different values of Purchase Expense from employing different filter constant values, the minimum value of Purchase Expense can be found. The corresponding mb and msc are the optimum sizing of the HEES, which allocate the capacity between lead-acid batteries and superca-

Y. Zhang et al. / Electrical Power and Energy Systems 65 (2015) 246–253

251

pacitors with lowest Purchase Expense. In this case, the minimum life cycle cost, which also includes the Other Expense containing ‘‘control and accessory cost’’ (Cca), ‘‘auxiliary and adjusting cost’’ (Caa), ‘‘recycle and environment spending’’ (Cre) and ‘‘operation and maintenance cost’’ (Com), is ¥8,116,896. The corresponding capacity allocation for lead-acid batteries are 4.90 MW, 1.27 MW h; for supercapacitors are 1.91 MW, 0.0229 MW h. Case 2 In this case, the defined generator profile Pd(t) is shown in Fig. 6, whose requirement of the wind power fluctuation leveling is more rigorous. The steps of determining the capacity allocation are similar to the steps taken in Case 1. The minimum life cycle cost in this case is ¥10,175,842. The corresponding capacity allocation for lead-acid batteries are 4.24 MW, 2.44 MW h; for supercapacitors are 1.93 MW, 0.0245 MW h. Fig. 6. Defined generator profile that requires medium level of smoothing.

Case 3 In this case, the defined generator profile Pd(t) is shown in Fig. 7, whose requirement of the wind power fluctuation leveling is very rigorous. The steps of determining the capacity allocation are similar to the steps taken in Case 1 and 2. The minimum life cycle cost in this case is ¥13,864,908. The corresponding capacity allocation for lead-acid batteries are 4.32 MW, 4.55 MW h; for supercapacitors are 1.94 MW, 0.0273 MW h. Demonstrating software platform development

Fig. 7. Defined generator profile that requires high level of smoothing.

Based on the developed algorithm, a programming software platform based on MATLABÒ (R2012a) language is developed. The schematic diagram is shown in Fig. 8. From left-side of the panel, the wind power historical data can be loaded by users from their stored Microsoft Excel files, by choosing ‘‘Wind Power (MW)’’ radio button in Historical Data of Wind Power Generation frame and clicking ‘‘Input’’ button when it is activated after the ‘‘Wind Power (MW)’’ radio button is chosen. From Choice of EES/Expense Input frame, the types of EES (which are Pb-acid battery and Supercapacitor) can be chosen by selecting

Fig. 8. Schematic diagram of the programming software platform.

252

Y. Zhang et al. / Electrical Power and Energy Systems 65 (2015) 246–253

Fig. 9. Employ programming software platform to demonstrate the sizing results of choosing different defined generator profiles: (a) requiring low level of smoothing; (b) requiring medium level of smoothing; and (c) requiring high level of smoothing.

Y. Zhang et al. / Electrical Power and Energy Systems 65 (2015) 246–253

corresponding checkboxes; and the unit prices of them can be typed into the textboxes when they are activated after the checkboxes are chosen. Other Expense, including ‘‘control and accessory cost’’, ‘‘auxiliary and adjusting cost’’, ‘‘recycle and environment spending’’ and ‘‘operation and maintenance cost’’, can be typed by users directly. From right-side of the panel, the defined generator profiles are pre-loaded into the program and can be chosen from the combo box in EES Scheduling Suggestions. The years of the projecting lasting and the deposit annual rate can be input into the Project Period (Year) and Deposit Annual Rate (%) input boxes for calculating the deposit return of the investment in the case of bank depositing. The Ragone plots for lead-acid batteries and supercapacitors can be pre-loaded into the program by users for their adopted EES from experimental power and energy data. As mentioned in Section ‘Case studies’, the Ragone plots are Eq-battery = 0.0015  Pq-battery3 + 0.093  Pq-battery2 – 2.3112  Pq-battery + 33.876 (0 < Pq-battery < 36.8) and Eq-supercapacitor = 0.0062  Pq-supercapacitor + 5.8 (0 < Pq-supercapacitor < 935.48) for the lead-acid batteries and supercapacitors respectively that are adopted in our work. Then, clicking of the ‘‘Programming’’ button will run the sizing program. The Fig. 9(a)–(c) show the sizing results of choosing three different defined generator profiles, as discussed in the case studies. Conclusions In this work, we have developed a sizing strategy for allocating the capacity between high energy density sources (e.g. lead-acid batteries) and high power density sources (e.g. supercapacitors). Within this strategy, the minimization of LCC is adopted as the objective function, which regards the employment of HEES for realizing wind power integration as an economic choice; and the Ragone plots of EES are introduced in as constraints, which take the mutual restrictions between the rate capability and storage capacity of the EES under consideration. The optimization process, accompanying with three different case studies, is analyzed and studied in detail. Finally, a programming software platform based on MATLAB language is constructed based on the developed sizing strategy; and the demonstrating results for the case studies are shown. Acknowledgement This work is supported by Knowledge Innovation Program of the Chinese Academy of Sciences (Grant No.: KGCX2-EW-330), China Postdoctoral Science Foundation (Grant No.: 2011M501033), and the National High Technology Research and Development Major Program of China (863 Program) (Grant No.: 2011AA05A113). References [1] Chu S, Majumdar A. Nature 2012;488:294–303. [2] Editorial, Nature 457 (2009) 357–357. [3] Cyranoski C. Nature 2012;488:372–4.

253

[4] Bindner H, Riso F. Power control for wind turbines in weak grids: concepts development. Risø National Laboratory; 1999. [5] Leadbetter J, Swan LG. J Power Sour 2012;216:376–86. [6] Dany G. In: IEEE Porto Power Tech Conference 10th – 13th September, Porto, Portugal; 2001. [7] Holttinen H et al. Wind Energy 2011;14:179–92. [8] Denholm P, Ela E, Kirby B, Milligan M. The role of energy storage with renewable electricity generation 2010. NREL/TP-6A2-47187. [9] Bennett P, Lieskovsky J, McBee B. In: Barnes FS, Levine JG, editors. Large energy storage systems handbook. CRC Press; 2011. p. 17–50. [10] Levine JG, Barnes FS. In: Barnes FS, Levine JG, editors. Large energy storage systems handbook. CRC Press; 2011. p. 1–16. [11] S´wierczynski M, Teodorescu R, Rasmussen CN, Rodriguez P, Vikelgaard H. IEEE international symposium on industrial electronics (ISIE) 2010:3749–56. [12] Black M, Strbac G. J Power Sour 2006;162:949–53. [13] Johnson JX, De Kleine R, Keoleian GA. Appl Energy 2014;124:377–88. [14] Kloess M, Zach K. Int J Electr Power Energy Syst 2014;59:111–22. [15] Beaudin M, Zareipour H, Schellenberglabe A, Rosehart W. Energy Sustain Dev 2010;14:302–14. [16] Ibrahim H, Ilinca A, Perron J. Renew Sustain Energy Rev 2008;12:1221–50. [17] Pickard WF, Shen AQ, Hansing NJ. Renew Sustain Energy Rev 2009;13:1934–45. [18] Díaz-González F, Sumper A, Gomis-Bellmunt O, Villafáfila-Robles R. Renew Sustain Energy Rev 2012;16:2154–71. [19] Sarrias-Mena R, Fernández-Ramírez LM, García-Vázquez CA, Jurado F. Renew Sustain Energy Rev 2014;34:194–207. [20] Yago K, Koshi K. Fuji Electr Rev 2006;52:99–106. [21] Y. Yoshida, W. Guohong, T. Minakawa, International Conference on Power System Technology (POWERCON), 2010, pp. 1-7. [22] Yoshimoto K, Nanahara T, Koshimizu G. In: Power Systems Conference and Exposition (PSCE); 2006, p. 1244–51. [23] Zhang G, Tang X, Qi Z. In: Power and Energy Engineering Conference (APPEEC), Asia-Pacific; 2010, p. 1–4. [24] Tankari MA, Camara MB, Dakyo B, Nichita C. IREE 2010;5:521–9. [25] Zhang G, Tang X, Qi Z. Automat Electr Power Syst 2011;35:24–8. 93 (in Chinese). [26] Wu G, Yoshida Y, Minakawa T. Energy Procedia 2011;12:975–85. [27] Mahamadou AT, Mamadou BC, Brayima D, Cristian N. COMPEL 2011;30:246–64. [28] Yu P, Zhao Y, Zhou W, Sun H, Quan C, Liu J, et al. Power Syst Protect Control 2011;39:35–40 (in Chinese). [29] Yu P, Zhou W, Sun H, Guo L, Sun F, Sui Y. Proc CSEE 2011;31:127–33 (in Chinese). [30] Li W, Joos G. In: Power Electronics Specialists Conference (PESC); 2008, p. 1762–68. [31] Arulampalam A, Barnes M, Jenkins N, Ekanayake JB. IET Gener Transm Distrib 2006;153:701–10. [32] Tankari MA, Camara MB, Dakyo B, Lefebvre G. IEEE Trans Sustain Energy 2013;4:414–24. [33] Lv HR, Wang HF, Gao F, Ma JH, Yin WJ, Dong J, Wu JC. In: Zhang CS, editor. Materials Science and Information Technology; 2012, p. 1380–1385. [34] Mendis N, Muttaqi KM, Perera S. Industry Applications Society Annual Meeting (IAS); 2012, p. 1–8. [35] Xing ZX, Zhang GF, Liu JS, Yao XJ. In: Li YG, Li Y, Pan WG, editors. Progress in Renewable and Sustainable Energy; 2013, p. 487–93. [36] Li DS, Lu FL, Lv Q, Shang L. In: 45th North American Power Symposium; 2013, p. 1–6. [37] Chen G, Bao ZJ, Yang Q, Yan WJ. In: 10th IEEE International Conference on Control and Automation (ICCA); 2013, p. 1874–78. [38] Martinez JA. In: IEEE Power and Energy Society General Meeting; 2011, p. 1–6. [39] Christen T, Ohler C. J Power Sour 2002;110:107–16. [40] Christen T, Carlen MW. J Power Sour 2000;91:210–6. [41] Irving MR, Song Y-H. Power Eng J 2000;14:245–54. [42] Clark CE. In: Ginley DS, Cahen D, editors. Fundamentals of Materials for Energy and Environmental Sustainability, Materials Research Society; 2012, p. 565– 75. [43] Paatero JV, Lund PD. Wind Energy 2005;8:421–41. [44] Nilsson JW, Riedel SA. Electric circuits. 6th ed. Upper Saddle River, N.J.: Prentice Hall; 2000. p. 277–286, 702–711. [45] T.M. Tham, Dealing with measurement noise - a gentle introduction to noise filtring; 1998, [accessed 26.06.14].